Adaptive loop bandwidth circuit for a PLL

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to phase-locked loop (PLL) circuits, and, more particularly, to adaptive loop bandwidth circuits for a PLL.

2. Description of the Related Art

A phase-locked loop (PLL) is a circuit that generates, or synthesizes, a periodic output signal that has a constant phase and frequency with respect to a periodic input signal. PLLs are widely used in many types of measurement, microprocessor, and communication applications. One type of phase-locked loop is the charge-pump PLL, which is described in Floyd M. Gardner, “Charge-Pump Phase-Lock Loops” <i>IEEE Trans. Commun</i>., vol. COM-28, pp. 1849-1858, November 1980, the teachings of which are incorporated herein by reference.

In a conventional charge-pump phase-locked loop, a phase detector (PD) compares the phase θ_INof the input signal to the phase θ_OUTof a feedback signal derived from the PLL output. Based on the comparison, the PD generates an error signal: either an UP signal (when θ_INleads θ_OUT) or a DOWN signal (when θ_OUTleads θ_IN), where the error signal indicates the magnitude of the difference between θ_INand θ_OUT. A charge pump generates an amount of charge equivalent to the error signal from the PD, where the sign of that charge indicates the direction of UP or DOWN. Depending on whether the error signal was an UP signal or a DOWN signal, the charge is either added to or subtracted from the one or more capacitors in a loop filter. The loop filter may have a relatively simple design, comprising a capacitor C_Sin parallel with the series combination of a resistor R and a relatively large capacitor C_L. As such, the loop filter operates as an integrator that accumulates the net charge from the charge pump. The resulting loop-filter voltage V_LFis applied to a voltage-controlled oscillator (VCO). A voltage-controlled oscillator is a device that generates a periodic output signal, whose frequency is a function of the VCO input voltage. Input and feedback dividers may be placed in the input and feedback paths, respectively, if the frequency of the output signal is to be either a fraction or a multiple of the frequency of the input signal.

In digital data applications in which a locally generated clock is synchronized to input data, one type of PLL is implemented with a PD that observes discrete periods, or snapshots, of the phase error between the input data and the clock. The clock phase correction is based solely on the polarity, or direction, of the phase offset. This type of PD is sometimes referred to as a “bang-bang” PD. A bang-bang PD samples the phase error of the input data in the following way: one sample is taken near the optimal sampling point, termed the center of the eye, and a sample is also taken at or near the transition point where the data switches to a new logic level, termed the edge transition. PLLs with bang-bang PDs are only nearly “locked” to the input data and form non-linear systems that make analysis difficult.

If the sampled data is different from the value sampled during the prior edge transition (termed a prior “edge transition sample”), then the edge transition sample is made before the data changes to a new value. In this case, the system clock is leading, and the PD generates a DOWN signal to decrease the speed, or frequency, of the clock. Similarly, if the sampled data is different from the next edge transition sample, then the system clock is lagging, and the PD generates an UP signal to increase the frequency of the clock. However, bang-bang PDs may introduce excessive jitter in the resulting clock signal since the clock is shrunk or expanded at every edge transition.

SUMMARY OF THE INVENTION

The present invention relates to a phase-locked loop (PLL) circuit that employs a phase detector (PD) generating an up/down signal based on the phase error between a data signal and a clock signal input to the phase detector. The PD senses excess jitter and extends the loop bandwidth to accommodate such excess jitter. Phase error is derived by sampling of the clock signal and at least one phase-shifted version of the clock signal by the data signal, and a retimed data is generated by the PD by sampling of the data signal by the clock signal. The sampled clocks are employed to generate a modified control signal with greater resolution in detecting the phase error, which, in turn, increases the loop bandwidth.

In accordance with exemplary embodiments of the present invention, a PLL includes a phase detector (PD), a first charge pump, and a voltage-controlled oscillator (VCO). The phase detector generates an up/down signal based on a phase difference between a data signal, a clock signal, and one or more phase-shifted clock signals input to the PD, wherein the phase difference is generated by sampling each clock signal by the data signal and a level of the up/down signal is based on the phase difference. The first charge pump generates a feedback voltage from the up/down signal. The VCO generates the clock signal based upon the feedback voltage.

BRIEF DESCRIPTION OF THE DRAWINGS

Other aspects, features, and advantages of the present invention will become more fully apparent from the following detailed description, the appended claims, and the accompanying drawings in which:

<figref idref="DRAWINGS">FIG. 1</figref> shows a block diagram of a phase-locked loop operating in accordance with an exemplary embodiment of the present invention;

<figref idref="DRAWINGS">FIG. 2</figref> shows a relation between the input data and the input clock of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref>;

<figref idref="DRAWINGS">FIG. 3</figref> shows the control signal generated by the phase detector of <figref idref="DRAWINGS">FIG. 1</figref>;

<figref idref="DRAWINGS">FIG. 4</figref> shows an exemplary embodiment of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref>;

<figref idref="DRAWINGS">FIG. 5</figref> shows the characteristics of a sampled non-return-to-zero charge pump;

<figref idref="DRAWINGS">FIG. 6</figref> shows the characteristics of a sampled return-to-zero charge pump that may be employed for the integral charge pump of <figref idref="DRAWINGS">FIG. 1</figref>;

<figref idref="DRAWINGS">FIG. 7</figref> shows a relation between the input data, the input clock, and a phase shifted version of the input clock for an alternative embodiment of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref> that modifies loop bandwidth in accordance with an exemplary embodiment of the present invention;

<figref idref="DRAWINGS">FIG. 8</figref> shows the control signal generated by an alternative embodiment of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref>;

<figref idref="DRAWINGS">FIG. 9</figref> shows an alternative embodiment of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref>; and

<figref idref="DRAWINGS">FIG. 10</figref> shows the control signal generated by a phase detector used for the PLL of <figref idref="DRAWINGS">FIG. 1</figref> when multiple input clock phases are employed to sample the input data.

DETAILED DESCRIPTION

<figref idref="DRAWINGS">FIG. 1</figref> shows a block diagram of phase-locked loop (PLL) <b>100</b> operating in accordance with an exemplary embodiment of the present invention. PLL <b>100</b> comprises voltage controlled oscillator (VCO) <b>101</b>, divide-by-N₁<b>102</b>, divide-by-N₂<b>122</b>, phase detector (PD) <b>103</b>, integral charge pump (ICP) <b>111</b> having charge pump <b>104</b> and capacitor C<b>1</b>, proportional charge pump (PCP) <b>112</b> having charge pump <b>105</b> and resistor R<b>1</b>, voltage adder <b>106</b>, and process/voltage/temperature (PVT) compensator <b>110</b> including amplifier <b>107</b> and capacitor C<b>2</b>. Though not explicitly shown in <figref idref="DRAWINGS">FIG. 1</figref>, charge pumps <b>104</b> and <b>105</b> are coupled to a common ground and biased with a reference voltage V_ref. PLL <b>100</b> tends to synchronize the frequency of the signal provided by VCO <b>101</b> to a frequency that is a multiple of the frequency of the input data DATA_in.

VCO <b>101</b> is shown as being implemented as an inductor-capacitor (LC) oscillator having a fixed inductor value and a variable capacitor value. The variable capacitor value may be implemented with two pairs of tunable varactors, each varactor pair having a capacitance tuned by an input voltage. One varactor pair sets the dominant, or “large” capacitance, while the other varactor pair sets the minor, or “low” capacitance. For the exemplary embodiment shown in <figref idref="DRAWINGS">FIG. 1</figref>, VCO <b>101</b> receives two control voltages, KL_vcoand KH_vco. As shown in <figref idref="DRAWINGS">FIG. 1</figref>, inductor <b>152</b> is in parallel with varactor pair VP_H<b>150</b> controlled by KH_vcoand varactor pair VP_L<b>151</b> controlled by KL_vco. Each of the two control voltages, KL_vcoand KH_vco, sets a capacitance of a corresponding varactor pair. The control voltage KH_vcoadjusts the large capacitance, while the control voltage KL_vcoadjusts the low capacitance. Other types of VCOs well-known in the art, such as crystal oscillator or ring-type VCOs, might be employed for VCO <b>101</b>.

Control voltage KH_vcois set by an outer feedback loop comprising PVT compensator <b>110</b> including amplifier <b>107</b> with transconductance gain gm and capacitor C<b>2</b>. PVT compensator <b>110</b> is employed to compensate for variations in operation caused by variations in process (such as semiconductor processes, voltage (such as in power supply), and temperature. Amplifier <b>107</b> receives a reference voltage V_refto set a nominal operating frequency of VCO <b>101</b>. In addition, amplifier <b>107</b> receives the voltage appearing across C<b>1</b>, which voltage is a function of the current I_icpfrom charge pump <b>104</b>. Effects of PVT variations of PLL <b>100</b> in the frequency of VCO <b>101</b> are exhibited at the voltage across C<b>1</b>, since the average DC voltage across C<b>1</b>. Variations in voltage across C<b>1</b> are adjusted by (gm/C<b>2</b>) by PVT compensator <b>110</b> and applied to VCO <b>101</b> so as to minimize effects of the variations. Control voltage KL_vcois set by an inner feedback loop as described subsequently.

Divide-by-N₂<b>122</b> divides the output signal frequency of VCO <b>101</b> by N₂, N₂being a positive number, to generate the clock CLK_in. For a bang-bang PD, CLK_inmay be generated so as to have a frequency that is twice the frequency of the input data DATA_in. Divide-by-N₂<b>122</b> may be implemented with a counter that counts by N₂. Other dividers, such as fractional dividers, may also be employed for divide-by-N₂<b>122</b>.

PD <b>103</b> is desirably implemented as a bang-bang phase detector, in which the input data DATA_inis employed to determine when to sample the input clock CLK_in. As shown in <figref idref="DRAWINGS">FIG. 2</figref>, for the described exemplary embodiment, the rising edge of DATA_insamples the clock CLK_infor detection of the phase difference Δφ between DATA_inand CLK_in. Based on the detected phase difference Δφ between DATA_inand CLK_in, PD <b>103</b> generates an UP signal when the frequency of CLK_inis too slow and generates a DOWN signal when the frequency of CLK_inis too fast, as shown in FIG. <b>3</b>. The falling edge of CLK_inis employed to sample the data value of DATA_into generate retimed output data DATA_out, as described subsequently with respect to FIG. <b>4</b>.

<figref idref="DRAWINGS">FIG. 4</figref> shows an exemplary embodiment of PD <b>103</b> of FIG. <b>1</b>. PD <b>103</b> comprises two D flip-flops <b>401</b> and <b>402</b>. D flip-flop <b>402</b> receives the input clock CLK_inand is clocked by the rising edge of input data DATA_in. Thus, D flip-flop <b>402</b> samples CLK_inwith DATA_ingenerate the UP/DOWN signal. D flip-flop <b>401</b> receives the input data DATA_inand is clocked by the rising edge of the complement of CLK_in. The rising edge of the complement of CLK_incorresponds to the falling edge of CLK_in, which complement of CLK_inmay be provided as the output clock CLK_outshown in FIG. <b>1</b>. Thus, D flip-flop <b>401</b> samples DATA_into generate re-timed output data DATA_out. One skilled in the art might employ other types of bang-bang phase detector known in the art for PD <b>103</b>.

Operation of the inner feedback loop to set control voltage KL_vcois now described. Returning to <figref idref="DRAWINGS">FIG. 1</figref>, the UP/DOWN signal from PD <b>103</b> is applied to ICP <b>111</b> and PCP <b>112</b> (multiplexer <b>109</b> is optional, as described subsequently, and for the following discussion the output of PD <b>103</b> is directly coupled to ICP <b>111</b> and PCP <b>112</b>). The UP/DOWN signal from PD <b>103</b> is provided as a proportional control signal PCS to charge pump <b>105</b> and is provided as an integral control signal ICS to charge pump <b>104</b>. The proportional control signal might be a non-return-to-zero (NRZ) signal, and the integral control signal might be a truncated version of the NRZ control signal, such as a return-to-zero (RZ) control signal.

Charge pump <b>104</b> steers current I_icpto capacitor C<b>1</b>, where charge pump <b>104</b> steers current I_icpup or down based on the integral control signal ICS from PD <b>103</b>. Similarly, charge pump <b>105</b> steers current I_pcpto resistor R<b>1</b>, where charge pump <b>105</b> steers current I_pcpup or down based on the proportional control signal PCS from PD <b>103</b>. Here, “steering” current refers to adding or subtracting charge with positive or negative current direction. Voltages across ICP <b>111</b> and PCP <b>112</b> (i.e., voltages V_intand V_propacross C<b>1</b> and R<b>1</b>, respectively) are combined by voltage adder <b>106</b> to generate by KL_vco. The voltage V_propsets the bang-bang frequency ±ΔF_bbof the inner loop, which are the bounds of the frequency variation of the inner loop. The voltage V_intsets the base, or center, frequency of the inner loop.

The voltages V_propand V_intmay be as given in equations (1) and (2): &lt;br /&gt;<i>V</i>_prop<i>=I</i>_pcp<i>*R</i><b>1</b>  (1)&lt;br /&gt;<i>V</i>_int=(<i>I</i>_icp<i>Δt/C</i><b>1</b>)  (2)&lt;br /&gt; where Δt is an increment of time based on the sampling period of PD <b>103</b> and is truncated for the integral path. An associated system stability ζ is as given in equation (3): <maths id="MATH-US-00001" num="00001"><math overflow="scroll"><mtable><mtr><mtd><mrow><mrow><mi>ζ</mi><mo>∝</mo><mfrac><msub><mi>V</mi><mi>prop</mi></msub><msub><mi>V</mi><mi>int</mi></msub></mfrac></mrow><mo>=</mo><mrow><mfrac><msub><mi>I</mi><mi>pcp</mi></msub><mrow><mo>(</mo><mrow><msub><mi>I</mi><mi>icp</mi></msub><mo>⁢</mo><mi>Δ</mi><mo>⁢</mo><mstyle><mtext> </mtext></mstyle><mo>⁢</mo><mrow><mi>t</mi><mo>/</mo><mi>C1</mi></mrow></mrow></mrow></mfrac><mo>=</mo><mrow><mfrac><msub><mi>I</mi><mi>pcp</mi></msub><mrow><msub><mi>I</mi><mi>icp</mi></msub><mo>⁢</mo><mi>Δ</mi><mo>⁢</mo><mstyle><mtext> </mtext></mstyle><mo>⁢</mo><mi>t</mi></mrow></mfrac><mo>⁢</mo><mi>R1C1</mi></mrow></mrow></mrow></mtd><mtd><mrow><mo>(</mo><mn>3</mn><mo>)</mo></mrow></mtd></mtr></mtable></math></maths>&lt;br /&gt; Consequently, the system stability ζ increases as the ratio of V_propto V_intincreases.

The combination of the two voltages V_propand V_intfrom voltage adder <b>106</b> drives VCO <b>101</b> into phaselock. For some embodiments of the present invention, C<b>1</b> might be formed in an integrated circuit (IC), and so the inner loop is desirably stable for all operating frequencies of PLL <b>100</b>. The inventors have determined that the higher the ratio of V_propto V_int, the more stable the inner loop of PLL <b>100</b>. If capacitor C<b>1</b> were an external component of an IC, capacitor C<b>1</b> might be set is large as is required for a given implementation in order to satisfy the stability requirement, but if capacitor C<b>1</b> is formed in an IC, limits might be imposed on the value of C<b>1</b>.

Employing ICP <b>111</b> and PCP <b>112</b> (i.e., employing separate charge pumps) allows for capacitance multiplication. Capacitance multiplication is a characteristic of a circuit that makes a capacitance value appear to be larger than the actual capacitance of a circuit component. Two charge pumps allow for preservation of the ratio V_propto V_intby reducing either i) I_icp, ii) Δt, or iii) a combination of these values, thus allowing for operation of PLL <b>100</b> that may otherwise require larger values of capacitance for C<b>1</b>.

A bang-bang phase detector is a non-linear system because the feedback phase error is non-linear, resulting in a non-linear loop, and so the phase response of the system has a slew-limited nature. Any applied jitter to an input signal of the system is tracked up to the stewing point of the loop. For a very small level of applied jitter, the loop will track the jitter so as to appear as a very wide-band loop. The 3-dB loop bandwidth F_3dBis inversely proportional to the applied jitter JT_in, as given in equation (4): &lt;br /&gt;F_3dB∝(ΔF_bb/JT_in)  (4)

Returning to <figref idref="DRAWINGS">FIG. 1</figref>, PLL <b>100</b> might operate during periods in which no data is present (i.e., when the signal DATA_indoesn't exist). Consequently, PLL <b>100</b> may monitor the output signal frequency of VCO <b>101</b> and to provide inner loop control voltages KH_vcoand KL_vcothat maintain the signal frequency of VCO <b>101</b> within a predefined set of frequency bounds. Optionally, PLL <b>100</b> further comprises phase-frequency detector (PFD) <b>108</b>, divide-by N₁<b>102</b>, frequency discriminator <b>120</b>, and MUX <b>109</b>.

Frequency discriminator <b>120</b> receives both CLK_inand the reference clock CLK_ref. For example, frequency discriminator <b>120</b> may monitor CLK_inand CLK_refand declare a loss of frequency lock when the difference between the input clock signals is greater than a predefined amount, such as 400 parts-per-million (PPM) in systems following SONET standards. Implementations of frequency discriminators for frequency discriminator <b>120</b> are well-known in the art. When frequency discriminator <b>120</b> declares a loss of synchronization, frequency discriminator <b>120</b> may generate a signal SWITCH to MUX <b>109</b> to switch the input of ICP <b>111</b> and PCP <b>112</b> from the output of PD <b>103</b> to the output of PFD <b>108</b>.

Divide-by N₁<b>102</b> divides the output of VCO <b>101</b> by N₁, where N₁is a positive number. PFD <b>108</b> operates as a phase detector in a similar manner to that described above with respect to PD <b>103</b>, and PFD <b>108</b> generates an up/down signal based on a detected phase error between the output of divide-by N₁<b>102</b> and a reference clock CLK_refat a predefined reference frequency.

The values of N₁and N₂for divide-by N₁<b>102</b> and divide-by N₂<b>102</b>, respectively, might be determined based on the particular implementation of PLL <b>100</b>. The value for N₁is set based on the ratio of VCO output signal frequency to reference clock frequency, and the value for N₂is set by the ratio of data signal frequency to reference clock frequency. For operation in accordance with the SONET standard, the reference clock frequency set as 155.52 MHz, and the VCO output signal frequency set as 2.48832 MHz, the values for N₁and N₂for varying data rates are as given in Table 1.

<tables id="TABLE-US-00001" num="00001"><table frame="none" colsep="0" rowsep="0"><tgroup align="left" colsep="0" rowsep="0" cols="4"><colspec colname="offset" colwidth="35pt" align="left" /><colspec colname="1" colwidth="84pt" align="left" /><colspec colname="2" colwidth="14pt" align="center" /><colspec colname="3" colwidth="84pt" align="center" /><thead><row>&lt;entry /&gt;<entry namest="offset" nameend="3" rowsep="1">TABLE 1</entry></row><row>&lt;entry /&gt;<entry namest="offset" nameend="3" align="center" rowsep="1" /></row><row>&lt;entry /&gt;<entry>Data Frequency</entry><entry>N₁</entry><entry>N₂</entry></row><row>&lt;entry /&gt;<entry namest="offset" nameend="3" align="center" rowsep="1" /></row></thead><tbody valign="top"><row>&lt;entry /&gt;</row></tbody></tgroup><tgroup align="left" colsep="0" rowsep="0" cols="4"><colspec colname="offset" colwidth="35pt" align="left" /><colspec colname="1" colwidth="84pt" align="left" /><colspec colname="2" colwidth="14pt" align="center" /><colspec colname="3" colwidth="84pt" align="char" char="." /><tbody valign="top"><row>&lt;entry /&gt;<entry>2.48832 MHz</entry><entry>16</entry><entry>1</entry></row><row>&lt;entry /&gt;<entry>622.08 MHz</entry><entry>16</entry><entry>4</entry></row><row>&lt;entry /&gt;<entry>155.52 MHz</entry><entry>16</entry><entry>16</entry></row><row>&lt;entry /&gt;<entry namest="offset" nameend="3" align="center" rowsep="1" /></row></tbody></tgroup></table></tables>

Some exemplary embodiments of the present invention might be employed for terminal equipment of a communications system operating in accordance with a SONET standard. For example, for a transmission terminal operating in accordance with a SONET standard, a recovered clock from a received signal is employed as the timing/synchronization source of the transmitter clock, hence a jitter transfer pole of that SONET standard is desirably adhered to. Such transmission terminal might employ an exemplary implementation of the present invention to generate a clock signal synchronized to input data received by the terminal. The specification for the jitter transfer pole of a SONET OC48 is 2 MHz. The JT_inbound is chosen by a tolerance mask value of the SONET standard as 0.15 UI_p-p(where p-p indicates peak-to-peak and UI indicates unit interval). From equation (4), the bang-bang frequency ΔF_bbwhere the onset of slew limiting occurs is 600 kHz.

For this example, the gain of the VCO should be relatively low, otherwise very small currents I_pcpand I_icp, or a small value of R<b>1</b>, is required to generate this 600 kHz bang-bang frequency. When the loop is locked, the generated jitter JT_outis desirably less than 0.1 UI_p-p. A value of KL_vcofor the given example might be selected as 40 MHz/V, if the total variation of the varactor pairs in the VCO is 500 MHz/V and the varactor pairs are a first pair having variation of 460 MHZ/V controlled by KH_vcoand a second pair having variation of 40 MHz/V controlled by KL_vco. For these pairs of varactors, R<b>1</b> is set as 1875 Ω and I_pcpis set as 5 μA. This gives a bang-bang frequency of 375 kHz, and a transfer bandwidth of 1.25 MHz.

Employing proportional and integral control signals to generate the voltages V_propto V_intin accordance with the present invention allows for capacitance multiplication, as described subsequently. For example, the stability of the system used for the SONET OC48 and OC12 standards might be analyzed using a behavioral model of the loop in the frequency/phase domain. Employing non-return-to-zero (NRZ) charge pumps for charge pumps <b>104</b> and <b>105</b> to receive the UP/DOWN signal from the bang-bang phase detector PD <b>103</b> (which is equivalent to using only proportional control signals for both charge pumps), and applying input data DATA_inas a length 2²³−1 pseudo-random bit sequence (PRBS), no instability might be observed. However, for the SONET OC3 standard, periods of instability might be observed.

To stabilize the loop of PLL <b>100</b>, such as for exemplary implementations for the SONET OC3 standard, the ratio of V_propto V_intis examined. If the loop is designed so that the UP/DOWN signals are NRZ, the loop is at either plus or minus the bang frequency until the next sample of phase difference is initiated. As shown in equation (3), the ratio of V_propto V_intindicates system stability, and the ratio of V_propto V_intis desirably maintained as much greater than 1.

From equation (3), as the Δt value decreases, the ratio of V_propto V_intincreases. The Δt value is inversely proportional to the data rate, since the Δt value is related to the sampling period between the input data DATA_inand input clock CLK_in. For an OC3 data rate this sampling period is 16 times longer than the sampling period for an OC48 data rate, as the baud rate is directly related to the NRZ time. To make the ratio of V_propto V_intthe same for both OC48 and OC3, as described previously, the charge pump of ICP <b>111</b> employs an integrated control signal ICS, and thus ICP <b>111</b> is selected as an RZ (return-to-zero) charge pump, limiting (reducing) the value of the At term in the relationship. Reducing the At term has the same effect as increasing (or “multiplying”) the capacitance C<b>1</b>.

<figref idref="DRAWINGS">FIGS. 5 and 6</figref> illustrate the difference between operation of charge pumps <b>111</b> and <b>112</b> when the integral and proportional control signals ICS and PCS are applied. <figref idref="DRAWINGS">FIG. 5</figref> shows the characteristics of the sampled NRZ charge pump that may be employed for PCP <b>112</b>, and <figref idref="DRAWINGS">FIG. 6</figref> shows the characteristics of the sampled RZ charge pump that may be employed for ICP <b>111</b>. For both <figref idref="DRAWINGS">FIGS. 5 and 6</figref>, the start and end of the correction cycle points are t₁and t₂. Thus, in <figref idref="DRAWINGS">FIG. 5</figref>, an NRZ control signal (across, for example, C<b>1</b>) yields a Δt_pterm for the voltage to change from V_INITto V_FIN. As shown in <figref idref="DRAWINGS">FIG. 6</figref>, the charge from the charge pump is delivered to the integration capacitor C<b>1</b> for a finite amount of time between t₁and t₂, which yields a Δt₁term for the voltage to change from V_INITto V_FINthat is much less than the Δt_pterm. Thus, employing the truncated proportional control signal instead of the proportional control signal over the entire correction cycle as in the NRZ charge pump of <figref idref="DRAWINGS">FIG. 5</figref> reduces the time that C<b>1</b> charges. Reducing the time that C<b>1</b> charges lowers the value of V_intthereby increasing the ratio of V_propto V_intto stabilize the loop.

A non-linear PLL exhibits a property that the bandwidth of the loop is proportional to the applied input jitter. However, setting the limit of the loop bandwidth based on a particular input jitter point, as described above with respect to the exemplary SONET loop stability, might not provide additional gain for excess jitter. Consequently, an alternative embodiment of the present invention employs a modified phase detector for PD <b>103</b> that senses excess jitter and extends the loop bandwidth to accommodate such excess jitter. Such modified phase detector generates an additional, phase-shifted version of the input clock which is also sampled by the input data. The two sampled clocks are employed to generate a modified UP/DOWN signal with greater resolution in detecting the phase error, which, in turn, increases the loop bandwidth by detecting and compensating for excess jitter more rapidly than the PD shown and described above for <figref idref="DRAWINGS">FIGS. 2</figref>, <b>3</b>, and <b>4</b>.

<figref idref="DRAWINGS">FIG. 7</figref> shows a relation between the input data DATA_IN, the input clock CLK_IN(shown as equivalent to the baseline or “in-phase” clock Iclk), and a 90-degree phase-shifted version of the input clock {overscore (Qclk)}. {overscore (Qclk)} is the complement of the 90-degree delayed (or “quadrature”) version Qclk of the input clock signal Iclk, and {overscore (Qclk)} leads Iclk by 90 degrees. Sampling of Iclk by the leading edge of the input data indicates whether the input clock CLK_INleads or lags the input data by at most 180 degrees of phase shift, and sampling of {overscore (Qclk)} by the input data indicates whether the leading or lagging input clock CLK_INis leading or lagging by more or less than 90 degrees of phase shift, allowing for an identification of the phase error between the data and input clock to within 0.25 UI of the data rate. <figref idref="DRAWINGS">FIG. 8</figref> shows the UP/DOWN control signal generated by an alternative embodiment of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref> that employs the in-phase and complemented quadrature phase of the input clock. As shown in <figref idref="DRAWINGS">FIG. 8</figref>, the magnitude of the UP/DOWN signal depends on the degree of phase, allowing the modified UP/DOWN signal to modify the currents I_icpand I_pcpaccordingly.

<figref idref="DRAWINGS">FIG. 9</figref> shows an alternative embodiment <b>900</b> of the phase detector of <figref idref="DRAWINGS">FIG. 1</figref> that may be employed to generate the UP/DOWN signal of <figref idref="DRAWINGS">FIG. 8</figref> in accordance with the exemplary clock signals of FIG. <b>7</b>. PD <b>900</b> comprises delay <b>901</b>, inverter <b>910</b>, D flip-flops <b>902</b>, <b>903</b>, and <b>904</b>, and adder <b>905</b>. CLK_INis provided as Iclk and is also provided to delay <b>901</b>. Delay <b>901</b> delays CLK_INby 90 degrees with respect to Iclk to provide Qclk. Qclk is applied to inverter <b>910</b>, which generates the signal {overscore (Qclk)} that leads Iclk by 90 degrees.

D flip-flop <b>902</b> generates an in-phase correct signal by sampling Iclk with DATA_IN. Similarly, D flip-flop <b>903</b> samples {overscore (Qclk)} with DATA_IN. The output of D flip-flop <b>903</b> is inverted and applied to D flip-flop <b>904</b> to sample the in-phase correct signal. When the rising edge of the input data signal DATA_INleads or lags Iclk, if the lead or lag passes the falling edge transition of {overscore (Qclk)} (which is the rising edge of the inverted output of D flip-flop <b>904</b>), then the phase error is greater than 0.25 UI. When phase error is greater than 0.25 UI, the output of D flip-flop <b>903</b> transitions to a logic high state, which passes in-phase correct signal through D flip-flop <b>904</b> as a quadrature phase correct signal. When phase error is less than 0.25 UI, the output of D flip-flop <b>903</b> transitions to a logic low state, which resets D flip-flop <b>904</b> and passes the logic low signal as the quadrature phase correct signal.

Adder <b>905</b> receives in-phase and quadrature correct signals. In-phase and quadrature correct signals are combined by adder <b>905</b> to form the UP/DOWN signal. When phase error is greater than 0.25 UI, the magnitude of the UP/DOWN signal is relatively large, and when phase error is less than 0.25 UI, the magnitude of the UP/DOWN signal is relatively small. The magnitude of the correction necessary to pull the PLL loop in lock is proportional to the magnitude of the UP/DOWN signal. Adder <b>905</b> may be implemented as a charge pump receiving the quadrature correct signal and coupled in parallel with ICP <b>111</b> and PCP <b>112</b>, which receive the in-phase correct signal.

One skilled in the art may extend the exemplary embodiment of <figref idref="DRAWINGS">FIGS. 7</figref>, <b>8</b>, and <b>9</b> to generate additional phase-shifted versions of the input clock signal to allow for greater accuracy in determining the phase error between the input clock and the input data. In addition, the delay between the different phase-shifted versions of the input clock need not be equivalent. Greater granularity might also allow for a variation in quantizing for the level of the UP/DOWN signal corresponding to the detected phase error. Thus, the greater the phase error, such as when a jitter “spike” occurs, an even greater level for the UP/DOWN signal might be employed to return the phase error to near lock faster. <figref idref="DRAWINGS">FIG. 10</figref> shows the control signal generated when i) multiple input clock phases are employed to sample the input data and ii) varying quantizing levels are employed for the UP/DOWN signal.

While the present invention has been described with respect to PLLs, the present invention is not so limited. The teachings of the present invention may be extended to delay-locked loops and similar applications. In addition, while the present invention has been described with respect to an LC oscillator, the present invention is not so limited. The teachings of the present invention may be extended to any oscillator, such as a crystal or ring-type oscillator, employed for a PLL or similar type of circuit.

The present invention may allow for the following advantages. A given implementation may allow for adaptive loop bandwidth of the loop filters in a PLL. In addition, the jitter exhibited by a given implementation may be maintained within a given specification of, for example, a SONET standard.

While the exemplary embodiments of the present invention have been described with respect to processes of circuits, the present invention is not so limited. As would be apparent to one skilled in the art, various functions of circuit elements may also be implemented in the digital domain as processing steps in a software program. Such software may be employed in, for example, a digital signal processor, micro-controller or general purpose computer.

The present invention can be embodied in the form of methods and apparatuses for practicing those methods. The present invention can also be embodied in the form of program code embodied in tangible media, such as floppy diskettes, CD-ROMs, hard drives, or any other machine-readable storage medium, wherein, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the invention. The present invention can also be embodied in the form of program code, for example, whether stored in a storage medium, loaded into and/or executed by a machine, or transmitted over some transmission medium, such as over electrical wiring or cabling, through fiber optics, or via electromagnetic radiation, wherein, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the invention. When implemented on a general-purpose processor, the program code segments combine with the processor to provide a unique device that operates analogously to specific logic circuits.

It will be further understood that various changes in the details, materials, and arrangements of the parts which have been described and illustrated in order to explain the nature of this invention may be made by those skilled in the art without departing from the principle and scope of the invention as expressed in the following claims.